Implications of teaching image processing without reference to computer vision

摘要

Two modules each lasting one semester, and running consecutively, are described in outline. The Image Processing module proceeds from local and global spatial domain techniques, convolution and Fourier transforms, to filtering and other frequency domain techniques. The Computer Vision module deals with binary image geometry and topology and pattern recognition (extended to neural nets). It is argued that combining the two into a single course could improve the learning experience.